[stella] Cycle-60 double-HMOVE magic

[Eckhard Stolberg]

>>
At least on my 7800, the sprites appear somewhat
distorted on the rightmost 32 pixels.  In particular, the sprite pattern
$A5 (X-X--X-X) sometimes appears as (X-X--XXX).  One of the missiles and
the ball are missing on every fourth scan line; the other missile appears
double every fourth scan line.  Note that because sprites move left once
every scan line while the kernel behavior is different on alternate scan
lines, the different behavior of the missiles is probably a result of
starting on an even or odd pixel.
<<

It seems to be more a question of when in the object output process
the extra HMOVE clocks occur. When you trigger the Cosmic Ark starfield
effect, you get a repeating pattern of two lines with missiles shifted
nomally by 17 pixels, one line with a double wide missile, and one
line with the missile missing.

Since you've been looking at the TIA schematics, did you find anything
in the object output circuits that could get affected like this by
the HMOVE pulses happening during the output?

And could you explain what triggering HMOVE during an ongoing
HMOVE process does exactly. It apparently resets the "more motion
needed" flags for all objects. But does it also affect the extended
HMOVE blank in any way?


Ciao, Eckhard Stolberg